The present invention relates to headend devices and more particularly to devices for providing high availability and fault tolerance in a cable system.
Cable communication systems typically route signals from a headend through trunk cables or fibers from which cables branch to individual users. The headend is the originating point in a communication system. Demand for connections through cable headends has greatly increased as cable usage has increased. Increased usage of cable modems and cable telephony is expected to place further demand upon headend operators. Cable headends therefore require high availability platforms having maximum fault tolerance for routing signals.
Providing a high availability platform in a cable headend environment is difficult. Typical cable headend units have multiple headend elements, each driving a separate cable wire. Physical limitation on the length of cable wire which can be driven by one headend element requires cable networks to be divided into smaller connection networks, each driven by a separate headend element. Even within these limits, cable headend units suffer high burnout, in part due to the power requirements and heat generated from driving a radio frequency (RF) signal. Failures often suddenly occur without prior indication, thereby causing an interruption in service. Typical radio frequency (RF) interfaces at a cable headend are switched using RF matrix switches which are well known to persons of ordinary skill in the art. Fault tolerance is achieved by using a matrix switch to transfer signals from a failed cable element to a good element. Any number of spare elements are typically provided for any set of cable elements that are attached to the matrix switch. A failed element is switched out when a fault is detected and an appropriate spare element is switched into its place while the fault is swapped out or repaired.
Fault tolerant systems have been developed to provide a single spare element which is capable of taking on the role of any one of a set of other elements in the cable headend. The number of spare elements and switching interfaces is reduced, thereby reducing cost and space requirements. Such fault tolerant systems are not scalable because they are typically switched using matrix switches which are available only in fixed N×N configurations. Incremental addition of a single spare element or a small number of spare elements in such systems may therefore require the expensive purchase and installation of a large matrix switch. Scalability is thus limited to installation of new element in blocks based on added switch matrixes.